Video and UGC
HeyGen Video 1.0 Review: Price per Second and Our Dutch Test
An AI clip for a cent per second sounds too good. We tested HeyGen Video 1.0 with Dutch scripts: what it does, what it costs and where it falls short.
Written by Genly's AI team, reviewed by Marlon Bonink
· 7 min read
AI video for one cent a second sounds like the answer for any small business that wants more video ads on a tight budget. Since HeyGen launched its own video model in late September, the question is whether it can replace a shoot or a creator. We are a Dutch company, so we also wanted to know how it handles a language that is not on its official list. We tested it ourselves on October 8, 2026.
Short answer: HeyGen Video 1.0 is a general-purpose AI video model from HeyGen, built on MiniMax H3, that generates 5 to 15 second clips with picture and sound in one pass, without an avatar. Through October 2026 it costs $0.01 per second at 480p and about $0.015 at 768p; after that, roughly double. In our own test, Dutch lines came out word for word, but the accent and pronunciation were not natural enough for Dutch talking-head ads. For B-roll, mood shots and product shots it is useful.
What is HeyGen Video 1.0?
You probably know HeyGen for its avatars: a digital presenter who reads your script. HeyGen Video 1.0 is different: a general-purpose video model that invents the whole scene, including the person, the setting, the light and the sound. Dialogue, sound effects and ambience come out of the same generation, with no avatar and no separate lip-sync step.
HeyGen built it on MiniMax H3, an open-weights video model, and post-trained it, according to HeyGen's documentation. It has been live since September 30, 2026 on the HeyGen API and through OpenRouter, Runware and ComfyUI. There was no regular web interface at launch: HeyGen describes access as "API-first". In practice, you either use it inside a tool that has built it in, or you need a developer.
The model has three modes:
| Mode | What you give it | Good for |
|---|---|---|
| Text to video | A written description (prompt) | Quickly testing an idea or a mood shot |
| Image to video | A photo that becomes the first frame | Bringing a product photo or a persona to life |
| Reference to video | Up to 9 images, 3 videos and 3 audio clips | Keeping the same person, product or voice across clips |
Clips run 5 to 15 seconds. Resolutions are 480p, 768p, 1080p and 2K; the last two were added in October, according to HeyGen's changelog. Vertical 9:16 for Reels, Stories and TikTok is supported.
HeyGen Video 1.0 pricing per second
HeyGen bills per second. Through the end of October 2026 there is a 50 percent launch discount: the product page lists pricing from $0.01 per second, down from a standard $0.02. At 768p the standard rate is $0.03 per second, and 1080p and 2K bill at three times the 768p rate. Tech site ProductNation ran a story on October 9 about AI clips from $0.01 a second. That is accurate, but only at 480p and only in October.
| Resolution | Launch price through October 2026 | Price after the sale | 10 second clip (sale) |
|---|---|---|---|
| 480p | $0.01 per second | $0.02 per second | $0.10 |
| 768p | about $0.015 per second | about $0.03 per second | about $0.15 |
| 1080p and 2K | 3x the 768p rate, about $0.045 | about $0.09 per second (our math) | about $0.45 |
In our own test on October 8, 2026, reference to video cost about $0.03 per second, roughly twice the price of text or image to video. Keeping the same person or product in every clip therefore costs around $0.30 per 10 seconds.
Our test: speed, consistency and Dutch speech
On October 8, 2026 we ran HeyGen Video 1.0 in all three modes, using one of our own AI personas.
Speed. A text or image to video clip was ready in 10 to 11 seconds. Reference to video took 4 to 8 minutes per clip.
Dutch speech. MiniMax H3 does not list Dutch among its eleven supported languages. Even so, Dutch sentences came out word for word. But the accent and pronunciation did not sound natural enough, and the quality varied from clip to clip. Our verdict: not (yet) suitable for talking Dutch UGC ads, where someone speaks straight to camera about your product. If you advertise in a language outside the official list, test before you commit.
Consistency. With our persona as a reference image, it was clearly the same person in 6 out of 6 clips. When we also added a voice sample as an audio reference, the voice sounded more consistent.
Strengths and weak spots:
- Useful for: footage without text (B-roll), mood shots and product shots.
- Weak at: long on-screen text, hair, paper and hand work. HeyGen says the same in its documentation.
- Watch out: sometimes music appeared under the clip even when the prompt said "no music".
HeyGen Video 1.0 vs Seedance 2.5, Kling and Veo
HeyGen publishes a blind comparison that puts HeyGen Video on top, ahead of Seedance 2.0, Kling 3.0 Pro and Veo 3.1, and says testers preferred it over MiniMax H3 Max in 55.8 percent of 650 blind votes. But this is HeyGen's own test, not an independent one, so focus on what you can verify: price, clip length and language.
| Model | Maker | Longest clip | Price per second with audio (API) | Dutch on the official language list? |
|---|---|---|---|---|
| HeyGen Video 1.0 | HeyGen, on MiniMax H3 | 15 seconds | $0.02 (480p) up to about $0.09 (2K), half that in October | No |
| Seedance 2.5 | ByteDance | 30 seconds | $0.10 (480p), $0.23 (720p), $0.57 (1080p) | No |
| Kling 3.0 | Kuaishou | 15 seconds | $0.126 (720p), $0.168 (1080p) | No |
| Veo 3.1 | 8 seconds | $0.40 (720p and 1080p), Fast from $0.10, Lite from $0.05 | No, prompt language is English |
Sources: price lists from BytePlus (Seedance, no video input), Kling and Google; language details from MiniMax H3, Seedance 2.5, Kling 3.0 and Veo 3.1.
The takeaway: at 480p and 768p, HeyGen Video 1.0 is the cheapest of the four per second. Seedance 2.5 makes clips up to 30 seconds and costs more. Kling 4.0 has been announced, but as of October 11 it was not yet on Kling's API price list. Veo 3.1 is the most expensive per second. And none of the four officially lists Dutch. We compare the tools that do handle Dutch in AI video generators for Dutch.
Frequently asked questions
Is HeyGen Video 1.0 the same as a HeyGen avatar?
No. An avatar has a fixed presenter speak your script. HeyGen Video 1.0 invents the whole scene, sound included.
Can I use HeyGen Video 1.0 in the HeyGen web app?
Not at launch. Access was API-first: the HeyGen API, OpenRouter, Runware or ComfyUI. Check with HeyGen whether a web interface has been added since.
How much does a 15 second clip cost?
During the October 2026 sale, about $0.15 at 480p and about $0.23 at 768p. After the sale, roughly double.
Can HeyGen Video 1.0 speak Dutch?
It can say Dutch sentences, even though Dutch is not on the MiniMax H3 language list. In our test on October 8, 2026, the accent and pronunciation were not natural enough and varied between clips. We do not (yet) recommend it for talking Dutch ads.
The bottom line
HeyGen Video 1.0 is cheap and fast for short scenes, which makes it a strong candidate for B-roll, mood shots and product shots. For an ad where someone talks about your product in Dutch, it is not good enough yet. That is why Genly picks the model that fits each video. Talking Dutch UGC videos are made in Genly with our own personas and Dutch voices. See how that works on our AI UGC page.
Editor
Marlon Bonink
Co-founder of Genly
This article was written by Genly's AI team and reviewed by Marlon before it went live. How we work with AI: AI transparency.
